ci(ruff): Set additional rules, code fixes (#353)

CHANGES

@@ -4,6 +4,10 @@
 
 <!-- Maintainers, insert changes / features for the next release here -->
 
+### Development
+
+- Ruff: Add additional rules, fix linting issues (#353)
+
 ## cihai 0.25.0 (2023-06-24)
 
 _Maintenance only, no bug fixes, or new features_

docs/conf.py

@@ -176,9 +176,7 @@
 autodoc_member_order = "groupwise"
 
 
-def linkcode_resolve(
-    domain: str, info: t.Dict[str, str]
-) -> t.Union[None, str]:  # NOQA: C901
+def linkcode_resolve(domain: str, info: t.Dict[str, str]) -> t.Union[None, str]:
     """
     Determine the URL corresponding to Python object
 

examples/variant_ts_difficulties.py

@@ -24,7 +24,7 @@ def run(unihan_options: t.Optional[t.Dict[str, object]] = None) -> None:
     print("3.7.1 bullet 4")
 
     for char in c.unihan.with_fields(["kTraditionalVariant", "kSimplifiedVariant"]):
-        print("Character: {}".format(char.char))
+        print(f"Character: {char.char}")
         trad = set(char.untagged_vars("kTraditionalVariant"))
         simp = set(char.untagged_vars("kSimplifiedVariant"))
         Unihan = c.sql.base.classes.Unihan
@@ -33,9 +33,9 @@ def run(unihan_options: t.Optional[t.Dict[str, object]] = None) -> None:
         else:
             print("Case 2 (non-idempotent)")
         for trad_var in trad:
-            print("s2t: {}".format(trad_var))
+            print(f"s2t: {trad_var}")
         for simp_var in simp:
-            print("t2s: {}".format(simp_var))
+            print(f"t2s: {simp_var}")
 
 
 if __name__ == "__main__":

examples/variants.py

@@ -7,7 +7,7 @@
 
 def variant_list(unihan: Unihan, field: str) -> None:
     for char in unihan.with_fields([field]):
-        print("Character: {}".format(char.char))
+        print(f"Character: {char.char}")
         for var in char.untagged_vars(field):
             print(var)
 

pyproject.toml

@@ -129,6 +129,26 @@ files = [
 ]
 
 [tool.ruff]
+target-version = "py38"
+select = [
+  "E", # pycodestyle
+  "F", # pyflakes
+  "I", # isort
+  "UP",  # pyupgrade
+  "B", # flake8-bugbear
+  "C4", # flake8-comprehensions
+  "Q", # flake8-quotes
+  "PTH", # flake8-use-pathlib
+  "ERA", # eradicate
+  "SIM",  # flake8-simplify
+  "TRY", # Trycertatops
+  "PERF", # Perflint
+  "RUF" # Ruff-specific rules
+]
+
+[tool.ruff.isort]
+known-first-party = ["unihan_etl", "cihai"]
+combine-as-imports = true
 
 [tool.ruff.per-file-ignores]
 "*/__init__.py" = ["F401"]

src/cihai/__init__.py

@@ -1 +1 @@
-from .__about__ import __version__ as __version__  # NOQA: F401
+from .__about__ import __version__ as __version__

src/cihai/_internal/config_reader.py

@@ -12,6 +12,16 @@
 FormatLiteral = t.Literal["json", "yaml"]
 
 
+class ConfigFormatNotImplementedError(NotImplementedError):
+    def __init__(self, format: str):
+        return super().__init__(f"{format} not supported in configuration")
+
+
+class ConfigExtensionNotImplementedError(NotImplementedError):
+    def __init__(self, ext: str, path: t.Union[str, pathlib.Path]):
+        return super().__init__(f"{ext} not supported in {path}")
+
+
 class ConfigReader:
     r"""Parse string data (YAML and JSON) into a dictionary.
 
@@ -46,7 +56,7 @@ def _load(format: "FormatLiteral", content: str) -> t.Dict[str, t.Any]:
         elif format == "json":
             return t.cast(t.Dict[str, t.Any], json.loads(content))
         else:
-            raise NotImplementedError(f"{format} not supported in configuration")
+            raise NotImplementedError(format=format)
 
     @classmethod
     def load(cls, format: "FormatLiteral", content: str) -> "ConfigReader":
@@ -102,14 +112,14 @@ def _from_file(cls, path: pathlib.Path) -> t.Dict[str, t.Any]:
         {'session_name': 'my session'}
         """
         assert isinstance(path, pathlib.Path)
-        content = open(path).read()
+        content = path.open().read()
 
         if path.suffix in [".yaml", ".yml"]:
             format: "FormatLiteral" = "yaml"
         elif path.suffix == ".json":
             format = "json"
         else:
-            raise NotImplementedError(f"{path.suffix} not supported in {path}")
+            raise ConfigExtensionNotImplementedError(ext=path.suffix, path=path)
 
         return cls._load(
             format=format,
@@ -184,7 +194,7 @@ def _dump(
                 indent=2,
             )
         else:
-            raise NotImplementedError(f"{format} not supported in config")
+            raise ConfigFormatNotImplementedError(format=format)
 
     def dump(self, format: "FormatLiteral", indent: int = 2, **kwargs: t.Any) -> str:
         r"""Dump via ConfigReader instance.

src/cihai/_internal/types.py

@@ -6,9 +6,9 @@
 :class:`StrPath` and :class:`StrOrBytesPath` is based on `typeshed's`_.
 
 .. _typeshed's: https://github.com/python/typeshed/blob/5df8de7/stdlib/_typeshed/__init__.pyi#L115-L118
-"""  # NOQA E501
-from os import PathLike
+"""
 import typing as t
+from os import PathLike
 
 if t.TYPE_CHECKING:
     from typing_extensions import TypeAlias

src/cihai/config.py

@@ -58,7 +58,9 @@ def expand_config(d: "UntypedDict", dirs: "AppDirs" = app_dirs) -> None:
         if isinstance(v, dict):
             expand_config(v, dirs)
         if isinstance(v, str):
-            d[k] = os.path.expanduser(os.path.expandvars(v).format(**context))
+            d[k] = os.path.expanduser(  # NOQA: PTH111
+                os.path.expandvars(v).format(**context),
+            )
 
             path = pathlib.Path(t.cast(str, d[k]))
             if path.exists() or any(

src/cihai/constants.py

@@ -26,5 +26,4 @@
 UNIHAN_CONFIG: "UntypedDict" = {
     "datasets": {"unihan": "cihai.data.unihan.dataset.Unihan"},
     # Turn off by default for using as a plugin example in examples/
-    # "plugins": {"unihan": {"variants": "cihai.data.unihan.dataset.UnihanVariants"}},
 }

src/cihai/core.py

@@ -1,12 +1,11 @@
 """Cihai core functionality."""
 import inspect
 import logging
-import os
 import pathlib
 import typing as t
+
 from cihai._internal.config_reader import ConfigReader
 from cihai.data.unihan.dataset import Unihan
-
 from unihan_etl.util import merge_dict
 
 from . import exc, extend
@@ -17,6 +16,7 @@
 
 if t.TYPE_CHECKING:
     from typing_extensions import TypeGuard
+
     from cihai.types import ConfigDict, UntypedDict
 
     DS = t.TypeVar("DS", bound=t.Type[extend.Dataset])
@@ -25,6 +25,11 @@
 log = logging.getLogger(__name__)
 
 
+class CihaiConfigError(exc.CihaiException):
+    def __init__(self) -> None:
+        return super().__init__("Invalid exception with configuration")
+
+
 def is_valid_config(config: "UntypedDict") -> "TypeGuard[ConfigDict]":
     return True
 
@@ -101,12 +106,14 @@ def __init__(
         expand_config(_config, app_dirs)
 
         if not is_valid_config(config=_config):
-            raise exc.CihaiException("Invalid exception with configuration")
+            raise CihaiConfigError()
 
         self.config = _config
 
-        if not os.path.exists(app_dirs.user_data_dir):
-            os.makedirs(app_dirs.user_data_dir)
+        user_data_dir = pathlib.Path(app_dirs.user_data_dir)
+
+        if not user_data_dir.exists():
+            user_data_dir.mkdir(parents=True)
 
         #: :class:`cihai.db.Database` : Database instance
         self.sql = Database(self.config)

src/cihai/data/unihan/bootstrap.py

@@ -1,7 +1,7 @@
 import typing as t
 
-import sqlalchemy.sql.schema
 import sqlalchemy
+import sqlalchemy.sql.schema
 from sqlalchemy import Column, String, Table
 
 from unihan_etl import core as unihan
@@ -40,19 +40,16 @@ def bootstrap_unihan(
 try:
     DEFAULT_FIELDS = [f for c, f in UNIHAN_MANIFEST.items() if c in ["Unihan"]]
 except Exception:
-    DEFAULT_FIELDS = [f for f in UNIHAN_MANIFEST.values()]
+    DEFAULT_FIELDS = list(UNIHAN_MANIFEST.values())
 
 
 def is_bootstrapped(metadata: sqlalchemy.sql.schema.MetaData) -> bool:
     """Return True if cihai is correctly bootstrapped."""
     fields = UNIHAN_FIELDS + DEFAULT_COLUMNS
-    if TABLE_NAME in metadata.tables.keys():
+    if TABLE_NAME in metadata.tables:
         table = metadata.tables[TABLE_NAME]
 
-        if set(fields) == {c.name for c in table.columns}:
-            return True
-        else:
-            return False
+        return set(fields) == {c.name for c in table.columns}
     else:
         return False
 

src/cihai/data/unihan/dataset.py

@@ -10,6 +10,7 @@
 
 if t.TYPE_CHECKING:
     from sqlalchemy.sql.schema import Table
+
     from ...conversion import ParsedVars, UntaggedVars
 
 

src/cihai/db.py

@@ -1,5 +1,6 @@
 """Cihai core functionality."""
 import typing as t
+
 from sqlalchemy import MetaData, create_engine
 from sqlalchemy.ext.automap import automap_base
 from sqlalchemy.orm import Session

src/cihai/extend.py

@@ -17,12 +17,13 @@
 from . import utils
 
 if t.TYPE_CHECKING:
-    from cihai.db import Database
     from sqlalchemy.engine import Engine
     from sqlalchemy.ext.automap import AutomapBase
     from sqlalchemy.orm.session import Session
     from sqlalchemy.sql.schema import MetaData
 
+    from cihai.db import Database
+
     DSP = t.TypeVar("DSP", bound=t.Type["DatasetPlugin"])
 
 
@@ -106,10 +107,7 @@ def add_plugin(
         namespace: str,
         bootstrap: bool = True,
     ) -> None:
-        if isinstance(_cls, str):
-            cls = utils.import_string(_cls)
-        else:
-            cls = _cls
+        cls = utils.import_string(_cls) if isinstance(_cls, str) else _cls
         setattr(self, namespace, cls())
         plugin = getattr(self, namespace)
 

src/cihai/log.py

@@ -119,7 +119,7 @@ def template(self, record: logging.LogRecord) -> str:
             Style.RESET_ALL,
             " ",
         ]
-        module_funcName = [Fore.GREEN, Style.BRIGHT, "%(module)s.%(funcName)s()"]
+        module_funcname = [Fore.GREEN, Style.BRIGHT, "%(module)s.%(funcName)s()"]
         lineno = [
             Fore.BLACK,
             Style.DIM,
@@ -131,7 +131,7 @@ def template(self, record: logging.LogRecord) -> str:
         ]
 
         tpl = "".join(
-            reset + levelname + asctime + name + module_funcName + lineno + reset
+            reset + levelname + asctime + name + module_funcname + lineno + reset
         )
 
         return tpl

src/cihai/types.py

@@ -5,6 +5,7 @@
 
 if t.TYPE_CHECKING:
     from typing_extensions import TypeAlias
+
     from cihai.extend import Dataset
 
 

tests/conftest.py

@@ -1,15 +1,12 @@
-import os
 import pathlib
 import typing as t
 import zipfile
 
 import pytest
-
 import sqlalchemy
 
 from cihai.data.unihan.constants import UNIHAN_FILES
 
-
 if t.TYPE_CHECKING:
     from .types import UnihanOptions
 
@@ -25,8 +22,8 @@ def fixture_path(tests_path: pathlib.Path) -> pathlib.Path:
 
 
 @pytest.fixture
-def test_config_file(fixture_path: str) -> str:
-    return os.path.join(fixture_path, "test_config.yml")
+def test_config_file(fixture_path: pathlib.Path) -> pathlib.Path:
+    return fixture_path / "test_config.yml"
 
 
 @pytest.fixture

tests/test_cihai.py

@@ -5,6 +5,7 @@
 
 """
 import typing as t
+
 import sqlalchemy
 
 import cihai
@@ -54,9 +55,11 @@ def test_unihan_options(
     )
     assert "Unihan" in app.sql.metadata.tables
     assert app.sql.metadata.tables["Unihan"].columns
-    assert set(app.sql.metadata.tables["Unihan"].columns.keys()) == set(
-        unihan_constants.UNIHAN_FIELDS + ["ucn", "char"]
-    )
+    assert set(app.sql.metadata.tables["Unihan"].columns.keys()) == {
+        *unihan_constants.UNIHAN_FIELDS,
+        "ucn",
+        "char",
+    }
     assert bootstrap.is_bootstrapped(app.sql.metadata)
 
 

tests/test_config.py

@@ -1,9 +1,11 @@
 import os
 import pathlib
 import typing as t
+
+import pytest
 from appdirs import AppDirs
+
 from cihai.config import expand_config
-import pytest
 
 if t.TYPE_CHECKING:
     from cihai.types import UntypedDict

tests/test_datasets.py

@@ -11,7 +11,6 @@
 import typing as t
 
 import pytest
-
 import sqlalchemy
 
 from cihai import conversion